SLC Catalyst: Learning in Exchange Systems: Human and Machine Interactions

George Mason University, Fairfax VA

Investigators

Abstract

The exchange of goods, services and information is pervasive in everyday life. In particular, interactions among individuals in society and workplaces are centered on exchange. Exchange systems are ancient in origin, universal across disparate human cultures, and-along with biological and ecological processes-are examples of self ordering complex systems. Two workshops will be held this year that will focus on learning and exchange in organizations, markets and society in general. Understanding and extending the capacity of the mind through exchange processes can have a tremendous impact on how individuals and organizations allocate their scarce resources and to can better utilize their resources to expand opportunities. These workshops will bringing together researchers who are developing new interdisciplinary algorithms and methodologies for artificial learning and those who are trying to understand aspects of learning as they apply to market design and coordination along with the neuroscience (cognitive and economics) associated with personal and impersonal exchange. The workshops will investigate the design and use of software agents, data visualization and machine learning to expand the scope of exchange systems. IN particular, the advent of globalization and the proliferation of distributed communication systems have made the design of exchange institutions a vital aspect of the economy. The transfer of information and the way it is presented can assist in distributed learning systems. In addition, the incentives provided by an institution's rules can affect the quality of the information provided in a distributed network and reinforcement learning from the decisions made in the network. Without an exchange institution that aggregates information and distributes it in a timely matter, the scarce resources of the brain would be used to try and find trading partners and information concerning value trade-offs. This idea of the market as an extension of mind can now be supplemented with a unique function in the specialization and division of learning. As the environments in which exchange occurs become more complex and distributed, the use of computer algorithms to assist humans in focusing their learning to specific large reward/small cost areas can facilitate exchange. These raise question concerning when a software agent responsible for a transaction. Can a software agent enforce a transaction? Can software agents coordinate among each other to find solutions that are somehow represented to their human counterparts?